Fin implementation cost: per-outcome pricing on Salesforce

The service director has seen the price and likes it. Ninety-nine cents when the agent resolves a conversation, nothing when it does not, and no seats. The finance team's first question is the right one: if the software is that simple to buy, what is the rest of the money for? The answer is that Fin's licence is only one of three lines. The second is the Service Cloud org it works inside, and the third is the work that decides how often Fin succeeds.

#What changed on 10 September 2026, and what did not

Salesforce's investor release was brief about the deal itself: "Salesforce today announced it has completed its acquisition of Fin, formerly Intercom, an industry-leading customer agent company." The release says Fin will sit in Salesforce's AI Labs and continue to serve its customers. It says nothing about changes to pricing or packaging.

As at October 2026, Fin's own pricing page still lists per-outcome pricing, and its Salesforce page still describes running Fin on top of an existing Service Cloud org. That is the product this page prices. If Salesforce later folds Fin into Agentforce packaging or changes its rates, the figures here will move, so treat the vendor page as the authority and check it on the day you sign.

There is one practical consequence of the acquisition worth noting now. Until it closed, buying Fin for a Salesforce org meant a second vendor, a second contract and a second support desk. Whether that changes, and when, is a question to put to your Salesforce account executive and to Fin before you commit to a multi-year term.

#What you pay Fin

These are list prices from Fin's pricing page, as at October 2026, in US dollars. Intercom billing is in US dollars only.

Two rows in that table need a caution. The Copilot and Pro prices appear in the page's questions about Intercom plans. Fin's Salesforce page confirms that Copilot is available inside Service Cloud, but ask Fin in writing whether the same prices apply to a Salesforce deployment before you budget for either.

The billing rules on the same page are what make the model work. Fin states: "If Fin can't answer or fails to perform an action, or if at any point the customer asks to speak to a human, you pay nothing." It also states that you are charged at most once per conversation, even if Fin answers several questions, and that the outcome is refunded if the conversation is later reopened. Spending limits and alerts are available, and the page offers a 90-day money-back guarantee.

An outcome is not only a resolution. Fin counts an outcome when the customer confirms the issue is resolved, when they ask for no more help after Fin's answer, or when Fin completes a configured workflow, which it calls a Procedure, including a handoff. That last point matters for forecasting. A Procedure that ends in a deliberate handoff to your team is still a paid outcome, while a conversation where the customer simply asks for a person is not.

#What you still pay Salesforce

Fin does not replace Service Cloud. It works inside it. Fin's Salesforce page puts it this way: "Fin connects to your Salesforce org, syncs your Knowledge articles and customer data, and follows the rules and Flows you already use." Conversations Fin cannot finish are handed back to your representatives in Salesforce, live or as a case in the queue, so those representatives still need Service Cloud licences.

Fin's page also says you do not need to move to a higher Salesforce tier to deploy it. The edition decision therefore rests on what your human team needs. As at October 2026, the Service Cloud pricing page lists:

For comparison, Salesforce's own customer-facing agent is sold per conversation or per action. The Agentforce pricing page lists Conversations at US$2 (AU$2.80) each and Flex Credits at US$500 (AU$700) per 100,000, with a standard action using 20 credits. The units differ, so the comparison is not as simple as ninety-nine cents against two dollars. Fin charges only for successful outcomes; Agentforce Conversations charge per conversation handled. Which is cheaper for you depends on your resolution rate and conversation length, and you should model both with your own volumes.

#How outcome pricing behaves on a real queue

The appeal of outcome pricing is that cost follows value. The forecasting challenge is that it also follows your content and your customers.

Your monthly Fin bill is roughly your conversation volume multiplied by the share Fin resolves or completes through a Procedure, at 99 cents each, subject to the minimum. Both inputs move. Volume rises with launches, outages and seasonal peaks. The resolution share rises as your Knowledge improves and falls when a new product launches with thin documentation. A good implementation pushes the share up, and that raises the Fin bill while cutting the human cost behind it. Finance should expect that and judge the result by total cost per contact, not by the Fin line alone.

The reopen refund deserves attention too. If customers routinely come back to the same conversation because an answer was wrong, the refund protects you from paying for false resolutions, but your representatives still pay the cost in handling time. Track reopen rates alongside the bill.

#What the implementation is made of

Fin has no setup fee, and connecting it to Salesforce is quick. The work that decides whether it pays off is everything around the connection.

Connection and data access. Connect Fin to the org and choose which Salesforce data it may read and write. Fin uses Contact, Account and Case attributes, and writes transcripts, summaries and approved field updates back. Decide which fields it may update, and make sure no Flow or other automation writes the same field.

Knowledge clean-up. Fin answers from your Knowledge articles and other content. Out-of-date, duplicated or contradictory articles produce wrong answers and reopened conversations. Most teams find this is the largest single work package, and it is the one that most directly moves the resolution share.

Case routing. Fin responds to cases assigned through your existing rules or Flows, from email or web forms. Decide which queues and case types Fin takes first, and which it never touches, such as complaints, legal matters or vulnerable customers.

Channel choice. For live chat, Fin Messenger handles conversations on your website or app and can hand off to a live representative or create a case. If you already run Salesforce's own messaging for web and app, decide which one owns live chat. Two agents on one channel is a design failure, not a feature.

Procedures and handoffs. Build the workflows Fin completes, such as order lookups or address changes, and design the handoff so the representative receives the transcript, summary and field updates without asking the customer to repeat anything.

Voice, if needed. Fin can connect to a telephony system, logging calls as cases in Salesforce. Voice is priced separately through sales and needs its own testing.

Testing, cutover and hypercare. Test against real historical questions before go-live, start with a share of traffic, and watch reopens and handoff quality closely for the first month.

#The first ninety days

The way Fin is priced suits a staged start, and the staging is worth planning rather than leaving to chance.

The monthly minimum of US$49 covers 50 outcomes. During a pilot on a single queue or a slice of web traffic, you may resolve fewer than that, in which case the minimum is the bill. That is a small sum, and it means the cost of a careful pilot is almost entirely your own team's time and the partner's work, not the licence.

The 90-day money-back guarantee lines up with a sensible pilot length. Use it. Agree success measures before you start: the resolution share on in-scope cases, the reopen rate, the quality of handoffs as judged by the representatives who receive them, and customer satisfaction on Fin-handled conversations. If those measures are not met by the end of the window, you have a contractual exit as well as a technical one.

Spending limits are the other tool. Set one from the first day so that a surge in volume, or a misconfigured routing rule that sends every case to Fin, cannot produce a surprise. Raise it deliberately as you widen the scope.

#What makes the setup cheap or expensive

The cheapest Fin projects share three conditions. The Knowledge base is current and owned by someone who reviews it. Case routing already runs through clear rules or Flows, so pointing a slice of cases at Fin is a small change. And the scope starts with high-volume, low-risk questions, such as order status, password resets and opening hours.

The expensive projects are the reverse. Knowledge is scattered across a help centre, internal wikis and old macros, and nobody owns it. Routing is a tangle of assignment rules that few people understand. And the business wants Fin to take complex, account-specific requests on day one, which means building and testing several Procedures that call other systems.

Integration beyond Salesforce is the other swing factor. If Fin needs live order, billing or shipping data to resolve a question, someone has to expose that data safely, either through Salesforce or directly. That work is priced like any other integration and often outweighs the Fin configuration itself.

#Fin, Agentforce, or both

Since the acquisition, the most common question is whether to run Fin, Agentforce or both. The answer depends less on price than on ownership.

If you run both, give each a clear territory, by channel, brand or case type, and make sure only one agent ever replies to a given customer conversation. Agents that both answer, or that both write to the same case fields, create exactly the confusion they were bought to remove. If you run only one, choose on fit: how well each handles your channels, how much of your workflow already lives in Flow and Agentforce actions, and which pricing unit matches your conversation shape.

#Getting comparable quotes, and what gets left out

Because Fin's licence has no setup fee, partner quotes are almost entirely services. Give each bidder your monthly conversation volume by channel, your current resolution or deflection rates, the number of Knowledge articles and when they were last reviewed, the case types in scope, and whether Salesforce messaging is already live. Ask for a quote split into the work packages above and a twelve-month forecast of Fin outcomes.

Items that are commonly missing:

• Knowledge remediation, priced as a separate package rather than buried in configuration.
• The outcome forecast, including what happens to the bill as the resolution share improves.
• Copilot and Pro, if your representatives or managers will use them, at their own list prices.
• Voice outcomes, priced via sales, if phone is in scope.
• A full sandbox for testing write-back and routing, which Core lists as an extra purchase.
• Representative licence changes, if handling volumes fall and you plan to reduce seats at renewal.
• Contract terms, given the acquisition. Ask how pricing and support will work at your renewal date.

#Where to go next

For the deal itself, see what happened to Fin. Fin, formerly Intercom, and Salesforce explains what the agent does and how it sits in a Service Cloud org. The licence line on its own is on Fin Salesforce licence cost. For the architecture, including which system owns each conversation, read integrating Fin with Salesforce. If you are moving off Intercom history, should I migrate from Fin covers what to inventory, and hire a Fin Salesforce partner gives you the questions to put to a firm.